Log expected exceptions triggering retries at non-error level

[ashleyheath]

I would like to have INFO level logging switched on for Procrastinate's worker.py in order to see the start/stop points for the worker itself, along with any errors and warnings about misconfiguration, etc..

However, worker.py currently logs all unhanded exceptions from a task at ERROR, which results in error-level logging for expected exceptions. This is not desirable where the procrastinate task is being used to make use of the retry mechanism e.g. to repeatedly attempt a call to a remote API that can fail intermittently. In this context., I only want the final exception to be logged as an error (as now that there will be no more retry attempts, it's truly a problem). Prior to that, if the retry strategy is triggered, I believe INFO would be more appropriate.

Would you be open to this change being made? (Am happy to put a PR together).

[ewjoachim]

This seems appropriate (as long as one logs is still is "error"). Happy to review a PR :)

[ashleyheath]

@ewjoachim, got round to it eventually 😅 - #845 for you to review at your convenience. Thanks!